As the amount of digital information and communication increases, timely and accurate data collection for software based evaluation and correlation becomes increasingly more difficult. This increase in difficulty arises from three main issues: First, the shear volume of the data being generated. Second, the nature of the desired information (generally unexpected) relative to the comparison set and third, the desired information lies in the noise of the overall data set being evaluated.
Existing high performance computer designs face the historical challenges of generating enormous amounts of heat, processing latencies measured in the tens of nanoseconds as information is transmitted between components and, poor general processing flexibility due to dedicated vector processing and highly, to embarrassingly, parallel machines that perform poorly in general processing problems.
To put these challenges into perspective, a Pentium™ microprocessor generates approximately 100 W of waste heat in operation. For an individual computer this is manageable by heat pipes and convection cooling, but when used in an IBM supercomputer requiring over 130,000 microprocessors, the waste heat, cooling and power systems become incredibly complex and large. Processing latency time is directly related to the resistance in metal conductors. At room temperature a nanosecond is approximately the time required for an electron to flow through 12 inches of copper wire, whereas at liquid nitrogen temperatures, a latency time of tens of nanoseconds implies electrical conduction paths in excess of 50 feet. Performing processing in a relatively lossless manner, would have dramatic impact on the utility of ultra computing.
One lossless approach to overcoming the challenges of operating temperatures, latency times and processing limitations is to perform the computing function with light. However, for the last three decades optical computing has been the perennial technology of tomorrow because of the persistent limitations imposed by the lack of programmability and the slow translation of electric data into the optical regime.
Known prior art includes the following eight U.S. patents:
U.S. Pat. No. 7,113,327 issued to Gu et al. on Sep. 26, 2006 which discloses a high power fiber chirped pulse amplification system utilizing telecom-type components.
U.S. Pat. No. 6,873,454 issued to Barty et al. on Mar. 29, 2005 which describes a hybrid chirped pulse amplification system wherein a short-pulse oscillator generates an oscillator pulse that is stretched to produce a stretched oscillator seed pulse. A pump laser generates a pump laser pulse and the stretched oscillator seed pulse and the pump laser pulse are directed into an optical parametric amplifier to produce output amplified signal pulse and an output unconverted pump pulse that are directed into a laser amplifier producing a laser amplifier output pulse that is compressed to produce a recompressed hybrid chirped pulse amplification pulse.
U.S. Pat. No. 6,739,728 issued to Erbert et al. on May 25, 2004 discloses an easily aligned, all-reflective, aberration-free pulse stretcher-compressor in a compact geometry. The stretcher-compressor device is a reflective multi-layer dielectric that can be utilized for high power chirped-pulse amplification material processing applications. A reflective grating element of the device is constructed: 1) to receive a beam for stretching of laser pulses in a beam stretcher beam path and 2) to also receive stretched amplified pulses to be compressed in a compressor beam path through the same (i.e., common) reflective multilayer dielectric diffraction grating. The stretched and compressed pulses are interleaved about the grating element to provide the desired number of passes in each respective beam path in order to achieve the desired results.
U.S. Pat. Nos. 6,208,458 and 6,181,463 both issued to Galvanauskas et al. on Mar. 27, 2001 and Jan. 30, 2001, respectively, describe quasi-phase-matched parametric chirped pulse amplification system that substantially reduces the required pump peak power and pump brightness, allowing exploitation of spatially-multimode and long duration pump pulses. It also removes restrictions on pump wavelength and amplification bandwidth. This allows substantial simplification in pump laser design for a high-energy PCPA system and, consequently, the construction of compact diode-pumped sources of high-energy ultrashort optical pulses. U.S. Pat. No. 6,198,568 also issued to Galvanauskas et al. on Mar. 6, 2001 discloses use of Chirped Quasi-phase-matched materials in a chirped pulse amplification system wherein the limitations on maximum pulse energies from a fiber-grating pulse compressor are circumvented by placing a chirped-period quasi-phase-matched (QPM) crystal after the fiber-grating pulse compressor.
U.S. Pat. No. 6,061,379 issued to Schoen on May 9, 2000 teaches techniques for producing plasma x-ray laser amplifiers, encompassing laser generated high density, micron-sized plasma columns, and microwave driven low density, large plasma volumes which provide the population inversions necessary for x-ray lasing to occur in the plasmas.
U.S. Pat. No. 5,960,016 issued to Perry et al. on Sep. 28, 1999 discloses An all-reflective pulse stretcher for laser systems employing chirped-pulse amplification enables on-axis use of the focusing mirror which results in ease of use, significantly decreased sensitivity to alignment and near aberration-free performance.
As the amount of digital information increases, emphasis shifts from data collection to timely and accurate data processing. What is needed is an integration of fundamental advancements in the generation and recognition of optical patterns to improve the speed of high performance computing and data mining by two orders of magnitude.